Prognosis of Polycystic Ovary Syndrome (PCOS) by its Symptoms and Rotterdam Criteria

Polycystic Ovary Syndrome (PCOS) is a disease which cause the hormonal disorder in women which in result enlarged ovary with small cysts appearing on its outer edges. This paper attentions on the prognosis of PCOS by classifying its symptoms based on its Rotterdam criteria with respect to the symptoms. To diagnose PCOS, patients need to undergo various unnecessary tests and questionary which not only consume time and cost but has psychological effect on female patients. Women diagnosed with PCOS may face an elevated likelihood of developing type 2 diabetes, cardiovascular issues, hypertension, and endometrial cancer. The exact reason behind the cause of PCOS is not known yet. There are three most widely used Rotterdam criteria for diagnosing PCOS and two out of three criteria must be positive for diagnosis of PCOS. The three criteria are - Oligomenorrhea that means irregular menstrual cycle or Amenorrhea that is absence of menstrual cycle for long period of time, Hyperandrogenism that is clinical or biochemical signs of androgen in the body, polycystic varies in the ultrasound. For this various machine learning algorithms are used for classification and prediction. The dataset is subjected to various classification techniques such as gradient boosting classification, logistic regression, and hybrid approaches like the combination of random forest and logistic regression. From result analysis it is found that the most optimal method among various machine learning algorithms, Random Forest Classifier gives best accuracy for prediction of PCOS. The ensemble methods in machine learning gives the best performance and accuracy in decisions as it combines the insights of multiple learning models in machine learning. There are a lot of features that are obtained from clinical and biological tests, among all the 8 most promising features are selected.